AI Engineer (ML Systems & Infrastructure)

About the Role

We are looking for exceptional AI Engineers to build the next generation of AI infrastructure and Machine Learning Systems (MLSys).

This role focuses on large-scale system infrastructure rather than model research. You will work on the core foundations that power large-scale AI training and inference systems, including Kubernetes cluster management, RDMA networking, unified KV Cache architecture, observability platforms, distributed systems, GPU orchestration, and CUDA kernel optimization.

You will collaborate closely with AI researchers, infrastructure architects, networking engineers, and platform teams to maximize the efficiency, scalability, and reliability of AI systems.

Key Responsibilities

AI Infrastructure & Kubernetes

Design, deploy, and operate large-scale Kubernetes-based AI infrastructure.

Develop cluster governance frameworks, scheduling policies, resource isolation, and multi-tenancy capabilities.

Build and optimize GPU orchestration platforms using Kubernetes, Slurm, Volcano, Kueue, Ray, and related technologies.

Improve cluster utilization, reliability, elasticity, and operational efficiency.

RDMA & High-Performance Networking

Design and optimize RDMA, InfiniBand, RoCE, and high-speed Ethernet fabrics for distributed AI workloads.

Optimize GPU-to-GPU and GPU-to-NIC communication paths.

Improve distributed communication efficiency for large-scale training and inference.

Analyze and eliminate networking bottlenecks across AI clusters.

Unified KV Cache & Distributed Memory Systems

Design and implement unified KV Cache architecture across:

GPU HBM

CPU Memory

RDMA-accessible Memory

NVMe SSD

Distributed Storage

Develop efficient KV Cache sharing, migration, offloading, and scheduling mechanisms.

Optimize latency and throughput for large-scale inference systems.

CUDA & System Performance Optimization

Develop and optimize CUDA kernels for training and inference workloads.

Profile and optimize GPU compute, memory, communication, and scheduling efficiency.

Contribute to low-level optimization of AI frameworks and inference engines.

Work on technologies such as FlashAttention, TensorRT, Triton, NCCL, CUTLASS, and custom operators.

Observability & Reliability

Build end-to-end observability platforms for AI infrastructure.

Design monitoring, logging, tracing, alerting, and troubleshooting frameworks.

Develop performance dashboards and SLO-driven operational systems.

Improve maintainability, debuggability, and operational excellence of AI platforms.

Automation & Platform Engineering

Build automation tools for deployment, provisioning, monitoring, and operations.

Develop Infrastructure-as-Code (IaC) solutions using Terraform, Ansible, and related tools.

Build CI/CD pipelines and engineering productivity platforms.

Improve platform scalability and operational efficiency.

Required Qualifications

Education

Bachelor's degree or above in Computer Science, Software Engineering, Electrical Engineering, or related fields.

Technical Skills

Strong software engineering and programming skills.

Excellent system design capability and strong engineering craftsmanship.

Strong coding standards and code quality awareness.

Strong sense of ownership, accountability, and execution.

System Fundamentals

Strong understanding of:

Operating Systems

Computer Networks

Distributed Systems

Data Structures and Algorithms

Linux Internals

Programming Languages

Proficiency in one or more of:

C++

Go

Python

Rust

AI Infrastructure Experience

Hands-on experience in one or more of:

Kubernetes

GPU Infrastructure

Distributed Systems

AI Infrastructure

HPC (High Performance Computing)

Cloud-Native Platforms

Networking Experience

Experience with:

RDMA

InfiniBand

RoCE/RoCEv2

GPUDirect

NCCL

UCX

High-Speed Ethernet

GPU & Performance Engineering

Experience with:

CUDA

GPU Performance Optimization

Multi-GPU Systems

Distributed Training

Distributed Inference

Preferred Qualifications

Experience building large-scale AI training or inference clusters.

Experience with vLLM, SGLang, TensorRT-LLM, Triton, DeepSpeed, Megatron-LM, Ray, or similar frameworks.

Experience with unified KV Cache systems, memory hierarchy optimization, or distributed storage systems.

Experience with Kubernetes GPU Operator and NVIDIA Network Operator.

Experience with Prometheus, Grafana, Loki, OpenTelemetry, and observability platforms.

Experience contributing to open-source projects such as: vLLM, FlashAttention, CUTLASS, TVM, MLIR, Triton, Kubernetes, NCCL

Experience working across AI Infrastructure, HPC, Networking, and Silicon Systems is highly desirable.